Source: WikipediaWhat if the result of every docking result was published online, with associated metadata link to protocol, scoring function/docking function, etc)? What if we could search across this entire space to try and come up with some conclusions? What if every docking result was an addressable resource?
Just the kinds of questions we need to ask ourselves if we ever want to understand how to bring web-scale to our science.
Technorati Tags: Docking, Web-scale computing, Collective Intelligence
8 Comments
Together with Jean-Claude Bradley I’ve been doing that – right now it’s on a very small scale and there’s no markup. It’s part of the Useful Chem project but the goal was that if anybody wanted to rerun the docking – everything (input SMILES, conformations, PDB file, prepped PDB file, results) is publicly avaiable.
In fact, sometime back we (at IU) had started doing it in a systematic manner – basically docking a druglike subset (1M compounds) from PubChem into all the entries in PDBBind.And store poses, score values etc in an RDBMS. Due to automation issues we haven’t gotten beyond 7 proteins, so that project is languishing at this point.
It would be a good idea. I have always thought that there should be some sort of PDB-”metadata” kind of website, where computational experiments done on whatever PDB structure could be associated to the original model… for example solvated systems, drug docking studies, mutagenesis….
Here is an example of what Rajarshi is talking about:
http://usefulchem.wikispaces.com/D-EXP014
It would certainly be helpful for others to try to replicate his results using just the information on that page. That way we could learn how to report the docking runs better.
We also have a blank discussion section on that experiment. If we get any feedback – good or bad – that is where it would go.
Links to other docking runs:
http://usefulchem.wikispaces.com/combiugi
It would be a good idea. I have always thought that there should be some sort of PDB-”metadata” kind of website, where computational experiments done on whatever PDB structure could be associated to the original model… for example solvated systems, drug docking studies, mutagenesis….
Here is an example of what Rajarshi is talking about:
http://usefulchem.wikispaces.com/D-EXP014
It would certainly be helpful for others to try to replicate his results using just the information on that page. That way we could learn how to report the docking runs better.
We also have a blank discussion section on that experiment. If we get any feedback – good or bad – that is where it would go.
Links to other docking runs:
http://usefulchem.wikispaces.com/combiugi
I’ve had this thought myself many times while working on ligand binding energetic studies of various enzymes and proteins. I’m starting to come to the conclusion that our current understanding of electrostatic interactions vis-a-vis biomacromolecular docking in the most general sense — and our lack of understanding of how specific phenomenon, such as the way the molecular dipole moment impacts binding affinity — would be reconciled sooner than anyone thought if such a meta-database existed, and inasmuch our capacity for intelligent drug design would be thrust to an entirely new level. But then again, something like this is at least twenty years off.
I think benchmarking data like “Directory of Useful Decoys” are going in the same direction
http://dud.docking.org/
Has anyone asked Antony Williams of ChemSpider if this might be possible in the future? I am wondering, if this is really useful, because the data sets will be very large and for each parameter set you would need another result set?
Anyway, I like the idea, as long as it is transparent and benchmarkable.
I think benchmarking data like “Directory of Useful Decoys” are going in the same direction
http://dud.docking.org/
Has anyone asked Antony Williams of ChemSpider if this might be possible in the future? I am wondering, if this is really useful, because the data sets will be very large and for each parameter set you would need another result set?
Anyway, I like the idea, as long as it is transparent and benchmarkable.
One Trackback
[...] all of this would be open. Deepak asked the question the other day, ‘What if the result of every docking result was published online?‘ and I have previously imagined a ‘Genbank of SAR data‘. A programme like this [...]